In most varieties of North American English, the t sound in the middle of water is not the same as the t sound at the beginning of top. The t in water is performed very quickly and the blockage of the air-flow lasts for a much shorter period of time than it does during the t in top. For the t in top, the tongue tip carefully approaches the alveolar ridge, touches it, stays there for a brief time, and leaves again. For the t in water, the tongue tip is thrown in a ballistic motion and only grazes the alveolar ridge in passing.
This kind of very fast t is called a tap. The IPA symbol for it is [ɾ] -- essentially an [r] without the serif at the top left-hand corner.
A narrow transcription of some words with taps would be:
water | [ˈwɑɾɹ̩] |
butter | [ˈbʌɾɹ̩] |
city | [ˈsɪɾi] |
gotta | [ˈɡɑɾə] |
In North American English, a tap will usually be used instead of a regular [t] when it comes between two vowels (including syllabic [ɹ̩]s) as long as the second vowel is unstressed. (Other dialects of English will usually use a regular [t] here. North Americans will occasionally use a regular [t] here as well, but usually only in exaggeratedly careful, hyper-correct speech.)
Most North American speakers will also use a tap instead of a [d] in the same environment:
leader | [ˈliɾɹ̩] |
bloody | [ˈblʌɾi] |
coda | [ˈkoɾə] |
kiddo | [ˈkɪɾo] |
If both t and d are pronounced as taps in the same environment, it might be possible for the difference between them to be lost. For example, butter and budder might end up being pronounced as [ˈbʌɾɹ̩], and latter and ladder might both be pronounced as [ˈlæɾɹ̩].
This is indeed what happens in many North American dialects. In other dialects, words like latter and ladder are still pronounced differently. One way of writing the difference is to put the diacritic for voicelessness (a small circle) underneath the tap symbol. The symbol [ɾ] normally represents a voiced sound (like [d]). With the diacritic, [ɾ̥] represents a voiceless sounds (like [t]).
Some dialects |
Other dialects |
|
latter | [ˈlæɾ̥ɹ̩] |
[ˈlæɾɹ̩] |
ladder | [ˈlæɾɹ̩] |
[ˈlæɾɹ̩] |
But vibration of the vocal folds is seldom the real difference for those dialects that pronounce latter and ladder differently. (A tap is a very short period of time to try to get your vocal folds to stop vibrating then to start vibrating them again for the following vowel.) But there are a number of other small phonetic differences that usually go along with voicing. For example, if you say cat and cad several times normally, you should notice that the vowel of cat has a higher pitch than the vowel of cad and that the vowel of cad is longer. Using IPA diacritics (an acute accent for high pitch, a grave accent for low pitch, and a cuneiform-like colon [ː] for length), we can give narrower transcriptions of cat and cad as:
cat | [kǽt] |
cad | [kæ̀ːd] |
Marking the same pitch and length differences the latter/ladder contrast in the dialects that make it, we can give the following narrow transcriptions:
Some dialects |
Other dialects |
|
latter | [ˈlǽɾ̥ɹ̩] |
[ˈlæ̀ːɾɹ̩] |
ladder | [ˈlæ̀ːɾɹ̩] |
[ˈlæ̀ːɾɹ̩] |
It is also possible for taps to be nasalized. A nasalized tap (together with a fully nasalized preceding vowel) is a common replacement for [nt] in faster speech:
slower, more careful speech |
faster, more causal speech |
|
twenty | [ˈtwɛnti] |
[ˈtwɛ̃ɾ̃i] |
Toronto | [təˈɹɑnto] |
[ˈtɹɑ̃ɾ̃ə] |